Hello, I'm wondering how I would do something that seemingly would be simple, but I guess I've just been doing something wrong unfortunately. I've been researching and experimenting for days, trying out a bunch of different techniques and tutorials, asking around, and I still can't get it right, so it's been quite frustrating. I guess I'll start first with the intended goal of what I'm trying to do (with mental ray of course):

1. Film a clip with a camera, live footage.
2. Track it with match moving software like Boujou or PFTrack, etc.
3. Import the tracked scene into Maya
4. Place objects into the scene
5. Separate into render layers such as diffuse, color, specular, shadow, AO, etc.
6. Batch render the image sequences and composite them in other software (After Effects for example)

The problem I'm stuck with is rendering out render layers for something like a shadow render layer, where I could specify what is seen. For example, if the scene was a sphere on the live-footage table, I'd want just the shadow from the sphere for this one layer. When I test render an entire master layer, this process works fine and I can single out just the shadows onto the table plane and have everything else transparent. When I batch render however, my shadow catcher plane will be there as well, and I don't know how I can get just the shadow from the object.

Before I just go on with the different methods I've tried and what I've been doing wrong, here would be a good spot to pause and just ask, is there a good way to do this? Have a shadow catcher that has no alpha for itself besides the shadows cast by the objects? And for the background from the image plane of the camera to also not render out?

Thank you!